This paper presents a path following controller of a surface vessel with a prescribed performance in the presence of input saturation and external disturbances. Based on the three degrees-of-freedom model of the surface vessel, the designed backstepping control scheme features three functional parts, namely, guidance, attitude control, and velocity control. To guarantee that the position errors are confined within the prescribed convergence rates and maximum overshoot, a performance constrained guidance law is formulated with an error transformed function. Command filters are incorporated in the control subsections to limit the magnitude of the virtual controls and simultaneously avoid arduous computations involving their time derivatives. Subsequently, auxiliary systems that are governed by smooth switching functions are developed in an unprecedented manner to compensate for the saturation constraints on actuators. Nonlinear disturbance observers are concurrently introduced to estimate the unknown external disturbances for increasing system's robustness. It is demonstrated that under the proposed control, the prescribed transient and steady tracking performance bounds are never violated, and all closed-loop signals remain uniformly ultimately bounded, despite the presence of input saturation and disturbances. Results from a comparative simulation study illustrate the effectiveness and advantages of the proposed method.

The combination of bioinspired morphing wing and tail made of artificial feathers improves the flight performance of drones.

Common vulnerabilities in typical intelligent cyber-physical systems such as unmanned aerial vehicles (UAVs) can be easily exploited by cyber attackers to cause serious accidents and harm. For successful UAV operations, security against cyber attacks is imperative. In this paper, we propose a modified sliding innovation sequences (MSIS) detector, based on the extended Kalman filter optimal state estimation, for a dynamic quadrotor system to detect cyber attacks inflicted on both its actuators and sensors in real time. These cyber attacks include random attacks, false data injection (FDI) attacks and denial-of-service (DoS) attacks. The MSIS detector computes the operator norm of the normalized innovation (residual) sequence within a sliding time window and triggers the alarm if the value is above the preset threshold. For a quadrotor undergoing rapid turns in a complex trajectory, the detector observes a reduced false alarm rate as compared to other state estimation-based detectors. To address the initial estimation error problem, we implement an iteration procedure to initiate and calibrate the detector. By evaluating the sample covariance of the normalized innovation sequence, the MSIS detector has the capability to isolate cyber attacks. Finally, simulation results of a quadrotor in a periodic, complex trajectory flight are provided to verify the effectiveness of the MSIS detection and isolation method.

Equipping drones with target search capabilities is highly desirable for applications in disaster rescue and smart warehouse delivery systems. Multiple intelligent drones that can collaborate with each other and maneuver among obstacles show more effectiveness in accomplishing tasks in a shorter amount of time. However, carrying out collaborative target search (CTS) without prior target information is extremely challenging, especially with a visual drone swarm. In this work, we propose a novel data-efficient deep reinforcement learning (DRL) approach called adaptive curriculum embedded multistage learning (ACEMSL) to address these challenges, mainly 3-D sparse reward space exploration with limited visual perception and collaborative behavior requirements. Specifically, we decompose the CTS task into several subtasks including individual obstacle avoidance, target search, and inter-agent collaboration, and progressively train the agents with multistage learning. Meanwhile, an adaptive embedded curriculum (AEC) is designed, where the task difficulty level (TDL) can be adaptively adjusted based on the success rate (SR) achieved in training. ACEMSL allows data-efficient training and individual-team reward allocation for the visual drone swarm. Furthermore, we deploy the trained model over a real visual drone swarm and perform CTS operations without fine-tuning. Extensive simulations and real-world flight tests validate the effectiveness and generalizability of ACEMSL. The project is available at https://github.com/NTU-UAVG/CTS-visual-drone-swarm.git.

Drones, as advanced cyber-physical systems (CPSs), are undergoing a transformative shift with the advent of vision-based learning, a field that is rapidly gaining prominence due to its profound impact on drone autonomy and functionality. Unlike existing task-specific surveys, this work offers a comprehensive overview of vision-based learning for drones, emphasizing its pivotal role in enhancing their operational capabilities across various scenarios. First, the fundamental principles of vision-based learning are elucidated, demonstrating how it significantly improves drones’ visual perception and decision-making processes. Vision-based control methods are then categorized into indirect, semidirect, and end-to-end approaches from the perception-control perspective. Various applications of vision-based drones with learning capabilities are further explored, ranging from single-agent systems to more complex multiagent and heterogeneous system scenarios, while highlighting the challenges and innovations characterizing each domain. Finally, open questions and potential solutions are discussed to guide future research and development in this dynamic and rapidly evolving field. With the growth of large language models (LLMs) and embodied intelligence, vision-based learning for drones provides a promising yet challenging road toward achieving artificial general intelligence (AGI) in the 3-D physical world.